Multiscale Modeling of Polymers: from Atomistic Structure to Macroscopic Properties

Computer simulations serve as pillars of material engineering and design. Multiscale modeling enables the simulation of molecular phenomena at various spatial and temporal scales, aiding in the establishment of structure-property relationships for larger and more complex systems. This work focuses on predicting the properties of various polymer materials by utilizing atomistic and chemically specific coarse-grained models that enhance computational efficiency while preserving the underlying structure and characteristics.

We incorporated atomistic simulations to investigate chain motion in poly(vinyl acetate) thin films on silica. By comparing the simulation results with experimental data, we developed a comprehensive understanding of the distinct regions and relaxation modes contributing to segmental dynamics in these films. We quantified the impact of free and attractive interfaces on film dynamics and derived the temperature dependence of each relaxation mode.

Vinyl-addition polynorbornenes are high-glass transition polymers with backbone stereochemistry that is challenging to determine experimentally. We conducted simulations of polynorbornene systems with varying backbone tacticities and alkyl side chains to study the effect of microstructure on conformational, thermodynamic, and dynamic properties. Initially, we derived a representative atomistic model and subsequently created a coarse-grained model that preserves the underlying chemical microstructure to analyze high molecular weight polymers. Our results demonstrated that racemic chains formed a helical structure, and an increase in the ratio of meso isomers led to more rigid chains with faster dynamics. The helical structure persisted in alkyl-substituted racemic chains, but increasing the side chain length reduces the difference in dynamics among different backbone tacticities.

Bottom-up coarse-graining of macromolecules is commonly achieved by matching structural order parameters. Here, we introduced the distribution of nearest neighbors as an additional multibody order parameter and demonstrated that the inverse-Monte Carlo method can overcome challenges associated with the parameterization of various correlated potentials. We applied this technique to polyisoprene melts as a prototype system, and demonstrated that incorporating nearest-neighbor potentials provides a straightforward approach to match both thermodynamic and conformational properties. We find that several temperature state points can be addressed by refining the forcefield accordingly. We examined the dynamics of different coarse-grained models and found a similar acceleration of dynamics in different models.